Automatic contrast enhancement of low-light images based on Naka-Rushton visual adaptation in non-sub-sampled shearlet domain

In this paper a non-sub-sampled shearlet transform model for image enhancement to improve image quality by maximizing the information content in the input image is proposed and implemented. This is inspired by two basic human visual features: i) visual adaptation, ii) local contrast enhancement of color image. We achieve the goal in different stages: In initial stage evolved upon the shearlet transformation of an image from image space to a low-pass and high-pass images in shearlet space and then to the visual response space. The visual adaptation is generally implemented with the Naka-Rushton equation, a rational transformation which allows to accurately compress the range of the input image. The later stage consists in a variational method able to perform local contrast enhancement with Weber-Fechner's law. After modifying the response values, the transformation can be reversed to produce the resulting image. The proposed method Naka-Rushton shearlet Image Enhancement (NRSIE) has been implemented on several color images. The results are compared with two other popular color image enhancement techniques along with visual analysis, detail and background variance. It has been found that NRSIE scales better results compared to two other popular methods.